In a recent feature for the Guardian, David Goldblatt looks at EURO20 teams and identifies clusters of different typologies of diverse teams from the racial diverse teams representing the colonial empire past to those reflecting contemporary refugee migration, contrasting these to the more “monotone” teams such as Scotland. With this piece I want to expand on Goldblatt’s idea of a diversity barometer for EURO20 in two different ways. The first is to question whether ethnicity is the most appropriate content when the concept of diversity is applied to team sport. The second is to suggest a more nuanced approach to the role of diversity for team performance. I will do so by analysing diversity in teams from Group D of EURO20 including England, Scotland, Croatia and Czech Republic.

The overlap of diversity and cohesion

One problem with Goldblatt’s diversity analysis is that it cuts out half of the teams taking part to the European competition. Notably, the twelve countries from old colonial powers or German speaking and Scandinavian Europe considered the most ‘diverse’ are also the wealthiest and most affluent. The other twelve not considered are mostly from Eastern Europe and sit at the bottom of income indicator tables. There are countries in Europe that do not have enough economic attractiveness and political stability (either as a result of past colonial empire or current ambitions for civic nationalism) to show as much degree of affluence and hyper-cosmopolitanism. Rather, they are countries of (footballing) emigration. To identify a more inclusive notion of diversity and one more appropriate to team-level performance than race and ethnicity, I refer to the work of my colleagues Balazs Vedres and David Stark. By studying activities such as jazz band play and video game design, Vedres and Stark demonstrate that what’s important to achieve creative success is not just diversity but the overlap of diversity and cohesion. They develop a ‘topological’ understanding of diversity by looking at the overlap of culturally and cognitively different groups within a team, where the cultural-cognitive structure of a team is shaped by its members’ histories. The conjecture in their study is that cognitive distance in overlapping groups leads to teams’ creative success. In teams with limited cognitive distance, there is a reduction of innovation which leads to conformity. The tension brought about by cognitive distance between groups within a team leads to distinctiveness and creativity.

A diversity barometer for EURO20

I have been trying for some time to apply this notion of diversity to football teams. Having worked on careers as a research construct in earlier studies on workforce development in the IT sector, I tried to apply a similar approach and understand the effect of players’ histories on how they work as a team. An early example of this process is the Data Study Group on football player pathways that I lead at the Alan Turing Institute. More recently with a group of my data science students I completed a small study that looked at overlaps of players’ nationality, age and present and past club membership over the last ten years in Premier League football teams. The study investigated whether there is a correlation between the performance of irregular teams such as Chelsea and Newcastle and the cognitive distance in their groups of players by contrast to teams with more consistency in (sometimes disappointing) results such as Tottenham and Arsenal. With data from the four teams in Group D, I now want to show how this new notion can apply as well to develop a diversity barometer for EURO20. Focussing on career data from Wikipedia of the expected starting 11 as selected by the Guardian’s Experts Network, I reassess some of the assumptions in Goldblatt’s analysis i.e. that England is a ‘remarkably diverse’ team while Scotland is more ‘monotone’. By including two Eastern European national teams in the analysis (Croatia and Czech Republic), I also show that the notion of diversity I propose is more inclusive and applies as well to the national teams of countries that were not considered in the analysis published by the Guardian.

Is England so remarkably diverse?

Diversity as we define it reflects the degree of exposure to different languages and cultures as apparent in the international club experience of players in the national squad. Cohesion results from having had common experience in the past and in the current team, both at club and at national team level. Taking into account these factors, Scotland looks the second most diverse team in Group D with a 5.8 diversity coefficient after Croatia that sits at the top with a remarkable 8.05 and before Czech Republic with 5.5. England is the least diverse team in the analysis of cognitive diversity, with a 1.2 coefficient. Let’s start from comparing England to Croatia. Croatia players currently play for nine clubs from seven different countries while England players are all from domestic teams (seven of them). All but one player in the Croatia team had spells with foreign clubs compared to England’s one. In their career, members of the Croatia team played for 33 non-domestic clubs from 11 different countries. Noting that the Guardian predicted starting eleven does not include Trippier, Sancho or Bellingham, the only English player with international club experience is Mason Mount, with one year on loan at Vitesse where he collected 29 appearances. Croatia players combined total appearances for foreign clubs amount to an astonishing 1,966. This is without counting Rakitic’s 485 who recently retired from international duty. Of the ten players in the Croatia team with international club experience, 6 played for two or more leagues outside domestic. Vraslijko and Vida in three and Perisic four: Switzerland, Belgium, Germany and Italy.

Scotland leads on diversity above England and Czech Republic

Scotland being a (footballing) emigration country as much as Croatia and Czech Republic comes out well on the diversity barometer. The experience of the Scotland players with non-domestic leagues is not as vast as Croatia but superior to Czech Republic especially in terms of appearances. Uniquely, all of the Scotland players have had experience in non-domestic leagues where their combined number of appearance is 1,862. Scotland’s result is closer to Croatia’s 1,966 than Czech Republic’s 1,381. The difference to the East European teams is in how non-domestic experience is spread internationally. All but one Scottish player (Hedry) have had international club experience only in England. Czech players have collectively played for seven different leagues beyond domestic: Italy, Turkey, UK, Germany, Russia, Switzerland and Spain. While Croatia players account for four more also including France, Russia, Austria, Scotland and Belgium. However, it is worth noting that Scotland players put together can account for 24 different non-domestic clubs played for, which is higher than Czech Republic players who achieved only 23 despite the wider international reach. Age range, i.e. the difference between oldest and youngest player captures the variability in the mix of youth and experience in the team. This is arguably another factor of cognitive diversity. Scotland is doing well with age range too as it is exactly the same as Croatia (12 years) and higher than England (10) and Czech (9). This data and the comparison with the now apparently much more “monotone” England speak volumes about the cognitive diversity of (football) emigration countries such as Scotland, Czech Republic and Croatia.

Is diversity conducive of footballing success?

However, some might argue that diversity in itself is not necessarily conducive of footballing success. No matter how able players are to quickly adapt to so many different environments, languages and playing styles, it can still be argued that playing abroad and for different clubs can affect chances to create a cohesive national team as coaches have difficulties scouting and keeping track of their players abroad. This is a rather weak objection not only because teams such as Croatia proved the contrary in World Cup 2018 when they were runner-up beating England in the semi-finals. It is also the nature of national team competitions to suggest that cognitive diversity is an important factor. Coaches of national teams generally work less often with their players, squads are not as curated as clubs and cannot buy the players that they want. Skills deriving from cognitive diversity such as improvisation and the ability to adapt are key to international tournaments even more than club football. Still, it can be conceded cohesion is an important element in football performance also at international tournaments. Denmark and Greece won the European Championship in 1992 and 2004 against the odds with little or no player with international club experience. They won because, as they say, they performed as “a unit”.

Factors indicating degree of team cohesion

Let’s go back to Group D teams and look at data that can indicate degree of cohesion. One first factor that I used is common past experience. Seven of Croatia starting eleven had a common past experience at Dinamo Zagreb. Slavia Praga is the common background for six Czech players as Celtic is for as many Scotland players. England’s strongest core, despite the comparatively minor international spread, is just three players from Chelsea (four if you include Chilwell who only joined this year). When considering national team career at youth level, nine players of Croatia starting eleven went through the youth set-up collecting a combined total of 33 years spent with U21 against 20 for England with 8 player who progressed from the U21 squad. According to this specific indicator, Scotland is at the bottom with only seven first team players that also had U21 experience, for a mere combined total of 17 years. Comparatively, Czech have 9 and a cumulative total U21 experience of 24 years. As well as having played together in the past, another measure of cohesion is surely how long players have been playing together in their current team, in this case the national senior team. Here the numbers for Croatia stand out even more. Current Croatia starting eleven produces a combined total of 565 caps compared to England’s 314. If one takes into account Rakitic’s additional 106 caps, it is more than double than England. This is also due to a remarkable difference in average age between the two teams. 28.18 year of age is the average for Croatia, while England is one of the youngest teams in the tournament with an average age of 25.45. For Scotland, national first team total caps are exactly the same as England and above Czech by 12.

Group D diversity ranking

To summarise, Croatia leads also on cohesion with a coefficient of 8.5. Czech Republic comes second with 6.02, before England (5.54) and Scotland (5.03). Cohesion results do not look too good for Scotland. Although one should factor in the limited opportunities Scotland players had for not qualifying for a major tournament since 1996, their average age of 27.18 – much higher than England – should have predicted a different outcome. Overall, after considering diversity and cohesion together, Scotland comes close third in Group D with 10.83 points after Czech Republic with 11.52. Croatia is clearly leading the table with a 17.4 coefficient. England sits at the bottom with an overall rate of 6.74. In conclusion, a note of how the proposed study could be improved in the future. By adopting a truly topological approach, it could look at the exact extent of the overlaps in player histories. For example, as opposed to looking at the combined total caps in youth or senior national teams, it could consider co-occurrence more specifically and weight events differently when two or more players appeared together in the same games. Similarly, cognitive distance could be assessed more accurately by looking at factors that make players’ experiences more dissonant, for example playing in countries speaking different languages or for teams in different tiers of the pyramid.

EURO20 Diversity Barometer / Data Science & Football by blogadmin is licensed under a Creative Commons Attribution CC BY 3.0